Skip to main content

Agents

Build AI-powered autonomous agents that can reason, use tools, and maintain state across complex multi-step workflows.

Overview​

The Agents package enables you to create sophisticated AI agents using LLMs (Large Language Models) that can autonomously execute tasks, call tools, and make decisions. These agents can maintain conversation history, manage artifacts, and orchestrate complex workflows with minimal human intervention.

Key Features​

  • LLM Agent - Create autonomous AI agents powered by language models
  • Tool Integration - Define custom tools that agents can call to perform actions
  • Session Management - Maintain state and conversation history across agent interactions
  • Artifact Storage - Save and load files and data produced by agents
  • Agent Orchestration - Combine multiple agents in sequential, parallel, or loop patterns
  • MCP Support - Integrate with Model Context Protocol for extended capabilities
  • Callback Handling - Implement human-in-the-loop workflows with approval steps

Available Nodes​

Agent Types​

  • LLM Agent - Main agent node that uses LLMs for reasoning and decision-making
  • Sequential Agent - Runs multiple agents one after another
  • Parallel Agent - Runs multiple agents simultaneously
  • Loop Agent - Repeats agent execution based on conditions

Session & State​

  • Create Session - Initialize a new agent session with memory
  • Load Session - Resume a previously created session
  • Delete Session - Clean up session data
  • Get State - Retrieve current agent state
  • Update State - Modify agent state variables

Tools & Callbacks​

  • Tool In / Tool Out - Define custom tools for agents to use
  • Callback In / Callback Out - Handle human approval or input requests
  • MCP Tool - Connect to MCP servers for additional capabilities

Artifacts​

  • Save Artifact - Store files or data produced by agents
  • Load Artifact - Retrieve previously saved artifacts
  • List Artifacts - Get all artifacts in a session
  • Delete Artifact - Remove stored artifacts

When to Use This Package​

  • Complex Decision Making: When automation requires reasoning about multiple options
  • Dynamic Workflows: When the exact steps aren't known in advance
  • Document Processing: Having AI analyze and extract information from documents
  • Customer Service Bots: Building intelligent chatbots that can take actions
  • Research Automation: Agents that can search, analyze, and summarize information
  • Code Generation: AI assistants that can write and modify code